基于电子商务评论的商家信誉维度构建
发布时间：2018-11-20 06:20
【摘要】：【目的】通过对电子商务评论文本的分析和处理,获取有效的商家信誉信息,从客观角度建立商家信誉维度体系。【方法】基于HNC理论的同行优先原理和文本挖掘方法提出改进的评论文本主题词抽取方法和主题词聚类算法,并进行类簇标签抽取及各类簇权重计算。【结果】生成商家信誉维度体系及各维度权重,以京东平台手机评论文本为实例,构建商家信誉维度体系,并对其进行评价,证明方法的可行性与有效性。【局限】受HNC词库不全的影响需手工生成一部分字词符号,在应用到更大规模的评论文本处理时可能会存在限制。【结论】利用本文提出的方法建立的商家信誉维度体系能够客观地反映出用户真正关心的商品指标。
[Abstract]:[objective] to obtain effective merchant credit information through the analysis and processing of electronic commerce comment texts, From the objective point of view, the dimension system of merchant reputation is established. [methods] based on the peer-first principle of HNC theory and the text mining method, an improved method of subject word extraction and clustering algorithm of comment text is proposed. And the cluster label extraction and all kinds of cluster weights are calculated. [results] the dimension system of merchant reputation and the weight of each dimension are generated. Taking the mobile phone comment text of JingDong platform as an example, the business reputation dimension system is constructed and evaluated. To prove the feasibility and effectiveness of the method. [limitations] affected by incomplete HNC lexicon, a part of the word symbols need to be generated manually. There may be some limitations in the application of this method to the larger comment text processing. [conclusion] the merchant reputation dimension system based on the method proposed in this paper can objectively reflect the commodity index that the user really cares about.
【作者单位】： 大连理工大学管理与经济学部;
【基金】：国家自然科学基金重点项目“社会化商务中参与者的信誉与信任机理及交易决策研究”(项目编号:71431002)的研究成果之一
【分类号】：TP391.1
，
本文编号：2344040
[Abstract]:[objective] to obtain effective merchant credit information through the analysis and processing of electronic commerce comment texts, From the objective point of view, the dimension system of merchant reputation is established. [methods] based on the peer-first principle of HNC theory and the text mining method, an improved method of subject word extraction and clustering algorithm of comment text is proposed. And the cluster label extraction and all kinds of cluster weights are calculated. [results] the dimension system of merchant reputation and the weight of each dimension are generated. Taking the mobile phone comment text of JingDong platform as an example, the business reputation dimension system is constructed and evaluated. To prove the feasibility and effectiveness of the method. [limitations] affected by incomplete HNC lexicon, a part of the word symbols need to be generated manually. There may be some limitations in the application of this method to the larger comment text processing. [conclusion] the merchant reputation dimension system based on the method proposed in this paper can objectively reflect the commodity index that the user really cares about.
【作者单位】： 大连理工大学管理与经济学部;
【基金】：国家自然科学基金重点项目“社会化商务中参与者的信誉与信任机理及交易决策研究”(项目编号:71431002)的研究成果之一
【分类号】：TP391.1
，
本文编号：2344040
本文链接：https://www.wllwen.com/jingjilunwen/dianzishangwulunwen/2344040.html
